[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[gmt-dev] QVT review paper and UMLX

Hi All

There is an interesting review of the 8 QVT submissions at
http://www.omg.org/cgi-bin/doc?ad/03-08-02 and it doesn't seem
to be in the members-only area.

UMLX, though obviously not mentioned would shape up pretty well, falling
directly in 2 out of 3 of the recommended performance spaces. Only
2 submissions get even 1 out of 3.

The one UMLX misses is a moot point. M:N transformation from either
source or target models is recommended. UMLX is the only syntax with M:N.
Target-driven seems to me to be misconceived, the source must exist to be
converted. But what should be supported by any transformation suite,
is an analysis of the PIM to determine the source, an analysis of the
PDM to determine the shape of the target, and then a semi/fully automated
determination of the transdormation sequence to perform the PIM to PDM
compliant PSM. So the planning is target-driven but only in so far as
the target is described by a source. Overall it is multi-source driven.

UMLX doesn't address the issues of reconciliation and incremental update.
I perceive these as activities pereformed by further transformatons and
enabled by a meaningful abstract syntax. An incremental update can be
derived by tramsformation of the forward transormation. Similarly a
reconcilation of tweaked outputs can be performed by transformations
derived from the forward transform. Reverse transforms can be derived for
that sub-set of transformations that can be reversed. The paper
seems to ignore the well established proofs in Graph Theory that not all
transformations are reversible, particularly those that involve many
to one conversion. (How can a Java to HTML transformation (Javadoc) be
reversible unless extra information is cached in the HTML or the source
remains part of the output file-set.?)

	Regards
			
		Ed Willink

------------------------------------------------------------------------
E.D.Willink,                             Email: mailto:EdWillink@xxxxxxx
Thales Research and Technology (UK) Ltd, Tel:  +44 118 923 8278 (direct)
Worton Drive,                            or  +44 118 986 8601 (ext 8278)
Worton Grange Business Park,             Fax:  +44 118 923 8399
Reading,   RG2 0SB
ENGLAND          http://www.computing.surrey.ac.uk/personal/pg/E.Willink
------------------------------------------------------------------------